This invention relates to the field of font reproduction. In particular, the invention relates to font reproduction in electronic documents generated from scanned images.
There is a significant move towards digitization of printed matter such as books and newspapers to produce electronic versions of the printed matter. Printed matter contains a wide collection of historical fonts. Each font has significant number of variations. One of the challenges is to create a synthetic text reproduction which has the look and feel of the original font. This text reproduction is important in various use cases, including: e-book creation, word-based recognition, word-spotting, fixing typographical errors in the original, etc.
The conventional solutions of this problem try to create true-type font and usually include a font editor. For good reproduction it may also be necessary to find an accurate location of each character in a line and the typical distance to the next character. These kinds of solutions are very hard to implement and require human intervention.